conversation a data structure containing the authentication type and
mechanism, and use that to dissect the credentials in subsequent bind
responses.
Call the bind request and response dissectors regardless of whether
we're building a protocol tree or not, so that we call the
authentication subdissectors.
"read_string()" doesn't return anything through the string-pointer
argument if the string is zero-length; handle those cases.
svn path=/trunk/; revision=6241
it's handed a pointer, which means that "epan_dissect_prime_dfilter()"
doesn't do so either; make that argument a "const dfilter_t *" in both
cases.
svn path=/trunk/; revision=6239
Currently Ethereal sets and uses a default directory for reading
and writing, but only in some places. This set of patches extends
the setting of the default directory to the -w option as well as
the -r option, and causes all file dialogs to use and set the
default consistently. (I haven't changed the
Preferences/Printing/File dialog, though, as that's a special
case.)
There's also a fix for a bug where Ethereal was issuing the
message "Ring buffer requested, but capture isn't being saved to
a permanent file" even though a file was specified with -w.
There also appear to be some other cleanups in his patch.
svn path=/trunk/; revision=6238
o Added support for v1 and v7
o Use proto_tree_add_item() instead of proto_tree_add_text() to
enable filtering. Thanks sahlberg AT optushome.com.au for
pointing that out.
svn path=/trunk/; revision=6236
ranges specified with a mask, as well as manufacturer OUIs. Match the
address range values, as well as MAC addresses and manufacturer OUIs,
when translating MAC addresses to names.
Have "make-manuf" read a file containing the well-known addresses and
append it to the list of OUIs.
svn path=/trunk/; revision=6234
ranges specified with a mask, as well as manufacturer OUIs. Match the
address range values, as well as MAC addresses and manufacturer OUIs,
when translating MAC addresses to names.
Have "make-manuf" read a file containing the well-known addresses and
append it to the list of OUIs.
svn path=/trunk/; revision=6233
for items that should run to the end of the tvbuff.
Make the next level dissector handle local to "dissect_spnego()".
Handle negTokenInit's with mechTokens, by, when processing the mechTypes
sequence in a negTokenInit, remembering the first MechType and telling
"dissect_spnego_mechTypes()"'s caller about it, and having that caller
use that mechType to dissect the mechToken.
svn path=/trunk/; revision=6232
length of the string value and "item_len" to the length of the entire
BER item, including the type header. As such, "val_len", not
"item_len", should be used when formatting the value, and "item_len",
not "val_len", should be used as the length when putting the item into
the protocol tree and when advancing the offset.
The kvno in an encryptedData item is unsigned; display it as such.
svn path=/trunk/; revision=6231
as an argument, and looks up that OID in the GSSAPI OID hash table.
Always use that routine to look up OIDs, so that we never use the result
of "format_oid()" as the key (as that doesn't necessarily work).
Make "gssapi_oids" static, as one should only look up GSSAPI
authentication mechanism OIDs with "gssapi_lookup_oid()".
In the SPNEGO dissector, free up the OID strings when we're done with
them, and don't advance the offset past the OID until after we put the
OID into the protocol tree.
svn path=/trunk/; revision=6228
in the gssapi_oids hash table; the keys are just text representations of
the numbers in the OID, but "format_oid()" can add a formatted
description of the OID to the end of the string it returns, which means
it won't match.
Use -1 rather than "tvb_length_remaining(tvb, 0)" as the length to say
"this item goes to the end of the tvbuff.
svn path=/trunk/; revision=6225
that the desired authentication type is, too.
Don't bother updating the offset after dissecting the desired
authentication type - we don't update it after dissecting the identity
or notification.
svn path=/trunk/; revision=6224
Gtk1 is still single threaded so if the tap extensions need to do something
time consuming or cpu intensive, then the main application will suffer.
It is better than nothing.
svn path=/trunk/; revision=6215
so that it would be possible to share an unmodified gtk2-rpcstat.c with
any future gtk1-ethereal support for the tap system.
Anyone reading this cvs entry:
Perhaps someone could just try to add a draw_all_listeners() to the place in
file.c where it has finished reading or rescanning all packets.
That might work though it wouldnt update in semi-real time but would
be simple and better than nothing.
svn path=/trunk/; revision=6193